Moessbauer and magnetic studies of a mixed-valence ferrimagnet

Oct 10, 1977 - ERICK G. WALTON,'" DAVID B. BROWN,*Ib HERBERT WONG,IC and WILLIAM MICHAEL REIFFIC. Received May 19, 1977. AIC70365H...
0 downloads 0 Views 897KB Size
Volume 16 Number 10

Inorganic Chemistry

October 1977

0 Copyrighr 1977 by rhe American Chemical Society

Contribution from the Departments of Chemistry, University of Vermont, Burlington, Vermont 05401, and Northeastern University, Boston, Massachusetts 021 15

Mossbauer and Magnetic Studies of a Mixed-Valence Ferrimagnet, Fe2F5*2H20 ERICK G. WALTON,'" DAVID B. BROWN,*Ib HERBERT WONG,ICand WILLIAM MICHAEL REIFFIC Received May 19, 1977

AIC70365H

The red, crystalline mixed-valence compound pentafluorodiiron(I1,III) dihydrate, Fe2F5.2H20,has been examined using Mossbauer spectroscopy and magnetic susceptibilities. At 48 K the material orders ferrimagnetically as the result of spin compensation of iron(II1) by iron(I1). Above this critical temperature, the observation of distinct quadrupole-split Mdssbauer absorptions for iron(I1) and iron(II1) sites demonstrates the nonequivalence of oxidation states and establishes Fe2F5.2H20 as a class I1 mixed-valence compound. In the temperature range 49-300 K, the magnetic susceptibility may be explained using the molecular field theory of ferrimagnetism. Long range order is observed below 48 K. This order is demonstrated magnetically through the observation of spontaneous magnetization, field-dependent susceptibilities,and a saturation moment corresponding to a formula unit spin of S = Mossbauer spectra below the critical temperature exhibit magnetic hyperfine structure consistent with the ferrimagnetic ordering. Hysteresis in the magnetic behavior suggests the possibility of a low-temperature magnetic phase change.

Introduction In 1958 Brauer and Eichner2 reported that the interaction of metallic iron with hot concentrated hydrofluoric acid gave rise to a yellow solid of composition Fe2F5.7H20. They further showed that this material could be dehydrated, at 100 OC, to a red trihydrate and that at 180 "C reversible dehydration to a blue-gray anhydrous mixed-valence iron fluoride, Fe2F5, occurred. This system has aroused interest for several reasons. First, the variety of colors exhibited by these materials suggests that the complexes span a range of mixed-valence behavior3 and thus provide a unique opportunity to examine variations in mixed-valence interactions with only minimal changes in empirical formulation. second, the fluoride ion is known to be a unique species which is particularly effectiveas a bridging ligand with transition metals, suggesting at least the Pssibility of long range cooperative effects4 Finally, the systems are amenable to investigation by a variety of techiques, including Mossbauer spectroscopy and the field and temperature dependence of magnetization. Through the use of magnetic susceptibilities and Mossbauer spectroscopy we have recently formulated the yellow heptahydrate as the ionic class I mixed-valence species [ F ~ (H20)62'] [FeFj(H20)2-] A conclusion has been reached by Sakai and Tominaga6 on the basis of X-ray Powder patterns and 19Fbroad-line NMR spectra* The red species described by Brauer and Eichner2 as the trihydrate is in fact a dihydrate, Fe2F5.2H20.7-9There has been relatively little characterization of this dihydrate, although Sakai and TominagaIo have reported Mijssbauer spectra at room ternperature and 80 K. While our work was in progress, a more detailed examination of the low-temperature Mossbauer spectra of this complex appeared.ll F~~the most part, both of these studies are in agreement with our results, and where appropriate, a comparison to these reports Will be included in this paper. On the basis of color alone, the dihydrate is expected to be a class I1 mixed-valence complex. In this paper we report the

*'

2425

results of our studies, using several physical techniques, on this red complex, Fe2F5'2H20. The structure of this material, determined by single-crystal x-ray diffraction, has recently been reported*'2 Experimental Section The preparation of crystalline Fe2F5.2H20 has been described previously.'2 Magnetic Measurements. Magnetic susceptibilities were measured independently a t the University of Vermont and Northeastern University by the Faraday method, using HgCo(NC% as ca1ibrant.I' The variabletemperature magnetic susceptibility measurements carried out at Northeastern University were made on a Faraday balance composed of a Cahn R G electrobak", a Varian Model 4000 electromagnet with 4-in. constant force pole caps and a Janis Super Vari-Temp cryostat Over the range 1.5-300 K for ten fields between 1.6 and 5 kG using quartz fiber suspensions and sample holders. Temperature measurement and control were typically of the order AO.01 K or better and were achieved using a Leeds-Northrup K-5 potentiometer and a Lake Shore Cryotronics Model DT-5OOC set point controller, respectively, in conjunction with a calibrated silicon temperature sensor diode, a 10-rA constant current source, and an uncalibrated gallium arsenide control diode. Final temperature equilibration and stability were continuously monitored on a Leeds-Northrup Speedomax-XL 600-mV recorder that was used to read the error signal of the calibrated silicon diode after cancellation by the K-5 potentiometer. Temperatures below 4.2 K were measured via the vapor pressure of helium using Wallace-Tiernan Models FA-160 and 61-050 absolute pressure gauges while pumping was precisely controlled with an L.J. Engineering Model 329 vacuum regulator valve. Temperatures below 78 K and to as low as 50 K were also achieved using liquid nitrogen by pumping (Welch 1397) to well below the triple point on solid nitrogen. Both the vapor pressure of nitrogen and a calibrated silicon diode were used to monitor the temperature in the region 49-78 K. An Few. Bell Model 610 gaussmeter with a transverse Hall probe was used for measurement of magnetic fields. Data obtained by the Faraday method were checked, and magnetization measurements from 0 to 45 kG were obtained, using the PAR Models 150 and 151 vibrating sample magnetometers at the University of North Carolina. The gallium arsenide diode used to measure temperatures exhibits field-dependent

2426 Inorganic Chemistry, Vol. 16, No. 10, 1977

Brown et al.

behavior above about 10 kG. Consequently, high-field measurements were generally performed at 4.2 K, where temperature control is facile. At lower temperatures, stability was maintained by monitoring the vapor pressure of helium. MGssbauer Spectra. Mossbauer spectra in the interval 20-295 K were obtained as previously described for Fe2FS.7H20.5Mossbauer parameters are believed to be reliable to k0.02 mm/s. Measurements at lower temperatures were obtained with a Janis Super Vari-Temp cryostat with calibrated silicon sensor and control diodes. Mossbauer spectra in external (longitudinal) fields were determined at the Francis Bitter National Magnet Laboratory using a niobium-tin superconducting solenoid in conjunction with a conventional constant acceleration spectrometer and a source of 100-mCi 57C0on rhodium

metal. Electronic Spectra. Diffuse reflectance spectra were recorded at the University of Oxford on a Unicam SP700 double beam spectrophotometer with a Unicam diffuse reflectance attachment. Samples were finely ground and the spectra measured with respect to a magnesium oxide standard. Results and Discussion In the yellow heptahydrate, Fe2Fj.7H20, there are 12 potential ligands for the two metal ions, and in consequence both iron(I1) and iron(II1) attain six-coordination with no necessity for bridging ligands. However, the loss of five molecules of water upon forming the dihydrate suggests coordinative unsaturation and, in consequence, the necessity for extensive bridging interactions. This has now been confirmed by the single-crystal x-ray structure determination reported recently.I2 The structure consists of vertex-sharing octahedra Fe"'Fs forming parallel zigzag chains, which in turn are cross-linked by t r ~ n s - F e " F ~ ( H ~with 0 ) ~vertex sharing in the equatorial fluorine plane. Bridging fluorine is suggested by the electronic spectrum of this material as well. The diffuse reflectance spectrum14 of Fe2Fj-7H20exhibits a broad band due to the Fe(H20)62' ion with maxima at 9500 and 11 200 cm-'. The diffuse reflectance spectrum of the dihydrate exhibits a band of similar shape, but with maxima at 7900 and 9600 cm-'. This red shift is consistent with coordination of Fe2+by fluoride in Fe2Fs.2H20,since fluoride lies below water in the spectrochemical series. Apparently, as water, which is coordinated to Fe2+in Fe2F5.7H20,is lost upon dehydration, the vacant coordination sites are filled by formation of fluoride bridge bonds. Since iron(I1) and iron(II1) are linked by fluoride bridges in Fe2F5.2H20,it is expected that there will be some interaction between them, e.g., that the material will be a Robin and Day class I1 ~ o m p l e x .This ~ expectation is verified by the diffuse reflectance spectra. In addition to the broad band near 9000 cm-', the spectrum of the dihydrate exhibits a maximum near 15000 cm-I. This maximum is assignable to the intervalence transfer band, thus establishing that Fe2F5.2H20 is a class I1 mixed-valence material. It should be emphasized here that the material which we have studied is not identical with those previously described. In previous reports on the dihydrate (or, as erroneously described in earlier reports, the trihydrate) the material was prepared by the thermal dehydration of the heptahydrate.2,8-'' We have, however, prepared the complex directly, by-passing the heptahydrate stage. This material is analytically correct for Fe2F5.2H20, and its room-temperature Mossbauer spectrum, magnetic susceptibility, and x-ray powder pattern appear to be comparable to those of the dihydrate prepared by thermal dehydration of the heptahydrate. However, the two forms of the dihydrate exhibit quite distinct thermal behavior.'j Although it is possible that the difference between these two forms can be ascribed totally to a difference in crystallinity, it appears more probable (based on TGA, DTGA, DSC, and kinetic parameters) that there is some more fundamental difference. We hope to be able to comment more fully about any differences at a later date. It should be

L&

0

Figure 1. Molar susceptibility (a), inverse susceptibility (v),and effective magnetic moment (m) for Fe2F5.2H20as a function of temperature. Data are obtained at an applied field, H = 1.0 kG, with the sample cooled in the field. emphasized that all measurements discussed here refer specifically to the crystalline dihydrate prepared directly, rather than by dehydration. Magnetic Susceptibilities. Figure 1 shows plots of magnetic susceptibility, reciprocal susceptibility, and effective magnetic moment vs. temperature for Fe2F5.2H20over the temperature range 0-260 K at a single applied magnetic field of 1.O kG. The form of these curves is not common for coordination complexes and deserves comment. First, at higher temperatures (ca. 150-250 K) x-' appears to be approximately linear with temperature, suggesting Curie-Weiss behavior. Furthermore, extrapolation of this linear portion to x-' = 0 leads to a large negative intercept, consistent with antiferromagnetic ordering. In contrast to this, at temperatures lower than perhaps 150 K the curve deviates markedly from linearity. As the temperature approaches 50 K there is a rapid increase in susceptibility such that x-l falls, effectively to zero, which is behavior typical of ferromagnetic ordering. However, the overall behavior is, in fact, typical of ferrimagnetic ordering, with antiferromagnetic interactions leading to incomplete spin cancellation and a net ferromagnetic alignment. Since the information attainable and the treatment of the data are different above and below the ferrimagnetic Neel point, we will discuss the results for these two regimes separately. (1) Magnetism above the Neel point. Although the plot of reciprocal susceptibility vs. temperature appears to be linear above about 150 K, more detailed examination indicates that there is in fact still curvature. Thus, the least-squares fit to Curie-Weiss law behavior in the temperature interval 150-200 K leads to an expression xm = 5.65/(T 95), whereas the expression calculated for the interval 200-250 K is xm = 7.17/(T 174). Higher temperature intervals lead to increasingly negative intercepts on the temperature axis, and it may be concluded that the susceptibility only approaches Curie-Weiss law behavior asymptotically at higher temperature. Such behavior is, of course, expected on the basis of the Although more molecular field theory of accurate and refined models, which take specific cognizance of the fact that behavior is dependent on exchange interactions, are available for the description of cooperative effects in antiferromagnetic and ferromagnetic materials, molecular field theory provides at least qualitatively accurate descriptions for these materials, and for ferrimagnetically ordered materials it is the only model which provides satisfactory agreement with experiment. In the molecular field approximation the actual field acting on a particular atom is assumed to be the sum of an applied magnetic field and the internal or "molecular" field

+

+

Inorganic Chemistry, Vol. 16, No. IO, I977 2427

Magnetic Studies of a Mixed-Valence Ferrimagnet /

I

I 3. 0.5

1

O

i

I

4 1

u

00 60

'I II I I

-

120

Figure 3. Spontaneous magnetization of Fe2F5.2H20,plotted as reduced magnetization ( 0 )vs. reduced temperature, assuming T, = L 48.5 K. The solid line is the Brillouin function for S = '/2. I

due to interactions with other atoms in the material. Thus, an iron(II1) atom is assumed to be subjected to an external field as well as a molecular field resulting from its interaction (electrostatic, not magnetic, in origin) with both its iron(I1) and iron(II1) neighbors. These assumptions lead to the following form for the high-temperature behavior of the susceptibility 1 ( T - T,)(T- T,') -= x C(T-0') where T,, T,l, and 0' are functions of the various molecular field coefficients, which may be related to exchange integrals. This equation describes a hyperbola in which the susceptibility asymptotically approaches Curie-Weiss behavior at high temperatures and becomes infinite at a characteristic temperature, the ferrimagnetic Neel point (T,). Figure 2 shows the best fit of the experimental data (which is field independent) to this equation, and various parameters are indicated. The ferrimagnetic molecular field theory thus serves as a good model for the high-temperature magnetic behavior of Fe2F5-2H20. Some deviation at low temperature is, of course, expected, since the susceptibility can never become infinite. At high temperatures ( T >> T,) this equation reduces to the Curie-Weiss limit

0 = T,

+ T,'

(2)

- 0'

(3)

For Fe2F5.2H20,the limiting high-temperature susceptibility is then given by x = 6.74/(T 179) (4)

+

The similarity of this expression to that obtained from a linear fit in the region 200-250 K suggests that above perhaps 250 K the complex has effectively attained its Curie-Weiss limit. The effective magnetic moment, calculated from peff = 2.282(C'I2) is 7.34 and may be compared to a value of 7.91 pB predicted for noninteracting high-spin iron(I1) and ironthat is / . l & = [(pFe2+)2

+

(/+eS+)2]"2

10

VTC

180

Figure 2. Reciprocal susceptibility (*) vs. temperature above the Neel point. The solid line is the best fit of eq 1 to the experimental data, with T, = 48.0, T,' = -200.9, 8' = 25.8, and C = 6.74. The solid line represents the Curie-Weiss limit, calculated from eq 2-4.

1/x = (T - O)/C where

05

(5 )

A related material, LiFezF6,has been shown" to order antiferromagnetically at 105 K, and for this species perf,calculated in the same fashion, is 7.32 pB. The negative value of 0 in x = C / ( T - e) implies a negative exchange integral

between Fe2+and Fe3+,e.g., antiferromagnetic interaction. This conclusion is consistent with the more quantitative results obtained from measurements of the saturation magnetization at low temperatures. The value of the ferrimagnetic Neel point, T, = 48 K, is in excellent agreement with the value of T, = 48.6 K obtained by Imbert et al.," from low-temperature Mossbauer studies. We have independently made these same Mossbauer measurements and arrived at the same result. (2) Low-Temperature Magnetism. Below the ferrimagnetic Neel point (in this case, 48.0 K) long range order dominates the magnetic behavior of ferrimagnetic materials. Several features of the low-temperature magnetic properties of Fe2F5.2H20should thus be at least qualitatively predictable. These include (a) field-dependent susceptibility, (b) spontaneous magnetization, and (c) a saturation magnetization at 0 K compatible with spin compensation of Fez+ by Fe3+. At any point in the temperature-field domain the observed magnetization is given by the difference between the individual sublattice magnetizations, e.g.

-hf~~z+I

kfobd = 1 h f ~ ~ 3 +

(6)

In the absence of detailed knowledge of the individual sublattice magnetizations (which cannot be obtained from bulk magnetic measurements) it is impossible to quantitatively correlate experiment and theory, except at saturation. However, the observed magnetic properties of Fe2F5.2H20are in at least .qualitative agreement with expectations, and our results are discussed briefly below. (a) Spontaneous Magnetization. Below the ferrimagnetic Neel point Fe2F5.2H20exhibits spontaneous magnetization in zero applied field. Figure 3 shows the variation of the reduced magnetization (M( Z')/M(O K)) as a function of the reduced temperature (7 = T / T , ) . The shape of the curve is at least qualitatively that of the Brillouin function, as expected. In the extreme case, where Fe2+-Fe2+ and Fe3+-Fe3+ interactions are negligible in comparison to Fe2+-Fe3+interactions, the curve should approximate the appropriate Brillouin functionz0with S = f 2. This function, shown as the solid line in Figure 3, appears to provide a reasonable approximation to the data. Calculated curves with higher values of s provide a poorer fit to the experimental data. The observation that the curve with S = '/2 provides a better fit is consistent with our interpretation that the ferrimagnetic behavior results from antiferromagnetic interactions between Fe2+and Fe3+. The shape of the magnetization curve approximates "Type Q ' behavior,I6 which is the most commonly observed form for ferrimagnetic materials. The Mossbauer work of Imbert et al." demonstrates a clear discontinuity in the value of the magnetic hyperfine field at 26.5 K. This discontinuity should be apparent in the

2428 Inorganic Chemistry, Vol. 16, No. 10, 1977

I

Brown et al.

I

;6000 m

. ..* 1 .**

-2 3 0I 0

X

I

.'.

5

20

1c

15

Fl E - D (kGauss)

.e

Figure 5. Magnetization at 4.2 K as a function of increasing magnetic field, following cooling in zero applied field.

.O

T("K)

Figure 4. Susceptibility vs. temperature below the Nee1 point: curve A (e), Faraday balance data, 2.23 kG; curve B (a), VSM data, 1.0 kG, data obtained following cooling in zero applied field; curve C (u), VSM data, 1.O kG, data obtained following cooling in 1.O-kG applied field.

magnetization curve as well and may perhaps be reflected in the slight inflection in the curve near 25 K. However, it is clear that the effect in the magnetization data is not nearly so significant as it is in the Mossbauer data, a difference which might be a result of our use of crystalline Fe2F5q2H20,rather than the material prepared by thermal dehydration of the heptahydrate. (b) Field-Dependent Susceptibilities. As expected for a magnetically condensed system, the susceptibility becomes field dependent below the critical temperature, the susceptibility decreasing with increasing applied field. Although this field dependence is expected, the specific behavior observed for Fe2F5.2H20is complicated and suggests the existence of a distinct magnetic phase(s) which is as yet not characterized. Figure 4 shows several sets of susceptibility vs. temperature data in the range 0-75 K. The exact form of these data is dependent on the specifics of the measurements and requires some comment. Curve A represents data obtained using the Faraday balance at an applied field of 2.23 kG. It may be seen that the susceptibility exhibits a broad maximum near 20 K. Curve B represents data obtained using the vibrating sample magnetometer (VSM) at an applied field of 1000 G. These measurements were obtained by first cooling the sample to helium temperatures in zero applied field, applying the 1000-G field, and measuring susceptibilities while warming the sample. Although differing in detail, the forms of these curves are comparable, and once again the susceptibility maximizes at low temperature. Curve C, which again represents VSM data obtained at 1000 G, differs from curve B in that the magnetic field was applied while the sample was being cooled. The data are identical when measured during either warm-up or cool-down operation. Significantly, there is no maximum in the susceptibility curve. Thus, there must exist some field-dependent magnetic phenomenon which is effective at low temperatures and which leads to the observed difference between susceptibilities obtained for samples cooled either in or out of the field. At fields above the apparent magnetic phase transition (vide infra) the susceptibility is independent of sample treatment. Thus, a sample cooled in zero applied field and measured at 15 000 G on the warm-up cycle provides a curve identical with that obtained upon cooling in the field. In this case, as for all experiments conducted after cooling in an applied field, no susceptibility maximum is observed.

(c) Saturation Magnetization. Figure 5 shows the variation in the magnetization, M , of Fe2F5.2H20(calculated as the product of the molar susceptibility and the applied field) as a function of field at 4.2 K. As can be seen, the initial smooth increase in magnetization with increasing field is followed by an abrupt, and apparently discontinuous, increase at 14.7 kG. Following this rapid increase the magnetization is nearly constant with increasing fields up to 45 kG indicating that saturation has been reached. There is a marked hysteresis to this transition, since upon reducing the field the magnetization drops only slowly. The values of the magnetization at fields higher than the transition field, and at lower fields after first passing through the transition and then reducing the field, are comparable to values obtained upon cooling the sample in those applied fields, and of course much larger than the values observed following cooling in zero applied field and then raising the field to the appropriate value. Although a detailed interpretation of these observations has not proven possible, e.g., some type of spin-flop, metamagnetic, or canting behavior, this behavior is in at least qualitative agreement with the field-dependent behavior discussed above. As expected from the field-dependent behavior, the values of the magnetization below the transition field depend upon the sample's magnetic history but are independent of it above the transition. Unexpectedly, however, both the field at which the transition occurs and also the magnitude of the discontinuity appear to depend upon sample history, the rapid increase occurring on different occasions within the interval 11-15 kG. In fact, there may be several transitions involved at ca. 15 kG which are not readily resolved in powder magnetization studies. The saturation magnetization, M , at 0 K may be related to a saturation moment, psat,by

(7) Since saturation appears to be obtained at fields above the transition and since this saturation moment is nearly independent of temperature below about 4 K, the magnetization at 1.75 K and 15 kG may be taken as a good approximation to the zero degree saturation magnetization. The observed value is psat= 1.2 pB. Since the saturation moment is the maximum component of the magnetic moment in the direction of the applied field, e.g. it is possible to relate the spin and magnetization. A positive exchange interaction between Fe2+ and Fe3+ (e.!., ferromagnetism) would lead to a "molecular" spin of S = / 2 , hence a value of pusat = 9 pB, assuming the free-electron g value of 2.0. By contrast, a negative exchange interaction between Fe2+ and Fe3+ (e.g., the ferrimagnetic behavior as claimed here) leads to a molecular spin S = 1/2, hence an expected value of pat = 1.O p ~ This . is in excellent accord with the experimental value psat= 1.2 pB. This provides strong evidence for the existence of ferrimagnetic ordering in Fe2F5-2H20.

Magnetic Studies of a Mixed-Valence Ferrimagnet

I

-i

Inorganic Chemistry, Vol. 16, No. 10, 1977 2429

1

*2 +4 V E L O C I T Y (mm/sec) 0

Figure 6. Room temperature Mossbauer spectrum of Fe2F5-2H20. The solid lines through the data points represent a best fit to the sum of the three Lorentzian peaks indicated.

S+,

A,,

6+3

A,,

Temp, K

0

+2 + 4

Figure 7. Mossbauer spectrum of Fe2F5.2H20at 5 5 K. The solid line through the data points represents a best fit to the sum of the four Lorentzian peaks indicated.

Table I. Reported Mossbauer Parameters for Fe,F,. 2H,O and LiFe,FAa Substance

-2

VELOCITY (mm/sec)

Ref

10 Fe,FS.2H,0 1.58 2.57 0.73 0.56 293 11 Fe,F:2H,Q 1.59 2.50 0.73 0.59 295 Fe;F;.2H;O 1.60 2.44 0.70 0.65 295 This work 23 LiFe, F 1.61 2.53 0.72 0.45 295 Fe,F;.iH,O 1.71 3.31 0.86 0.59 80 10 50 11 Fe2F,.2H,O 1.73 3.30 0.78 0.58 55 Thiswork Fe,FS~2H,0 1.71 3.28 0.81 0.58 78 23 1.76 3.10 0.83 0.40 LiFe,F, a Allvalues for isomer shifts ( 5 ) and quadrupole splittings (A) are in mm/s, and isomer shifts are corrected relative to a sodium nitroprusside standard.

Mossbauer Spectra. Confirmation of pentafluorodiiron(I1,III) dihydrate as a class I1 rather than a class I11 substance3 is provided unambiguously by the Mossbauer spectra obtained above 50 K. The spectrum at 295 K (Figure 6) shows three distinct lines. The only reasonable assignment of Mossbauer parameters requires the assumption of overlapping Fe(I1) and Fe(II1) components of separate quadrupole split doublets. According to this view10y"*21two distinct quadrupole-split doublets exist, with parameters which are indicative of "trapped" valences-i.e., distinguishable Fe(I1) and Fe(II1) sites on the Mossbauer time scale. The best supporting evidence for the peak assignments is the spectrum obtained at 57 K (Figure 7 ) which clearly shows all four lines. The isomer shifts and quadrupole splittings of both Fe(I1) and Fe(II1) sites are similar to reported values for iron fluoride compounds22 and, in particular, to the mixed-valence salt LiFe2F623(Table I). For the iron(II1) sites, with Fe1"F6 coordination, the room temperature isomer shift (0.70 mm/s) is close to that of K3FeF6 (0.68 mm/s). The iron(II1) quadrupole splitting (0.65 mm/s) is surprisingly large for a material only slightly distorted from octahedral symmetry, and is, in fact, larger than in the [FeF5(H20)l2-ion.5 Although no other ion with the trans-Fe11F4(H20)2 configuration is known, the isomer shift (1.60 mm/s) observed for iron(I1) in Fe2F5-2H20is comparable to values in other ionic fluorides, and the quadrupole splitting (2.44 mm/s) is close to the values observed in FeC12.2H20 and FeBr2.2H20(2.50, 2.49 mm/s), both of which have the t r ~ n s - F e X ~ ( Hstructure. ~0)~ Although there does appear to be general agreement between our findings and previous reports10J1*21 of high-temperature

Mossbauer studies of the iron fluoride dihydrate, a closer examination of the tabulated results reveals a significant variation in certain of the parameters, particularly at room temperature. In all of the published spectra, as well as in our own work, the room temperature Mossbauer spectrum consists of three lines. The more intense peak at low velocity is assumed to result from an accidental degeneracy (in velocity) of the low-velocity branches of quadrupole split Fez+and Fe3+ spectra. This degeneracy must be nearly exact, since the absorption exhibits no broadening and since attempts to fit this peak as the superimposition of two peaks lead to no significant improvement in the fit and to identical velocities for both components. Under these conditions, that same peak velocity is used in the calculation of isomer shifts and quadrupole splittings for both Fe2+and Fe3+. Conversely, that peak position should be calculable from reported isomer shifts and quadrupole splittings according to velocity = 8+,, - AJ2

(9)

and should yield identical values of the peak velocity for both n = 2 and n = 3. However, examination of the data from ref 10 and 11 shows that the peak positions calculated from the Fe2+ and Fe3+ parameters differ by 0.14 and 0.10 mm/s, respectively. Consequently, either these authors were able to resolve the individual components of this absorption-a resolution which appears incompatible with published information-or else the calculations are in error. We suspect that this is the case and is the factor most likely to account for the discrepancy between published values and our work. Some of the differences in reported parameters may also arise in the modes of preparation. We have employed a novel preparation of the dihydrate which appears to give a highly stable, reproducible product which analyzes correctly as Fe2F5.2H20. All preparations of Fe2F5.2H20reported previously have followed a method suggested by Brauer,' i.e., thermal dehydration of Fe2F5-7H20at 100 OC. Although Sakai and Tominaga21have published thermogravimetric data which support the observation of a smooth, continuous dehydration from heptahydrate to dihydrate, we believe that the process is, in fact, complext5and that cdre must be taken in the preparation so as to avoid loss of H F and a consequent

2430 Inorganic Chemistry, Vol. 16, No. 10, 1977

Brown et al.

Figure 8. Mossbauer spectrum of Fe2Fj.2H20at 20 K The solid line through the data points represents a best fit to the sum of 11 Lorentzian lines. Individual Fe2+and Fe3+subspectra are labeled with stick diagrams.

F deficiency in the product.24 It is quite possible that the compositional integrity of Fe2F5.2H20may be dependent upon its mode of preparation. Although this is as yet unproved, it seems proper to enlist care in the comparison and interpretation of results from various laboratories. The Mossbauer spectra of Fe2FS.2H20at 20 K, Ho = 0, and at 4.2 K, Ho = 0, and longitudinal fields of 5 and 60 kG are shown in Figures 8 and 9, respectively. The ferrimagnetic ordering of the material is reflected in the hyperfine splitting at 20 K where two subspectra (Fez+ and Fe3+) are apparent and labeled with stick diagrams. The primed transitions correspond to the ferric sublattice and the double prime to the ferrous. The transitions of the ferric sublattice are sufficiently resolved so that one may obtain accurate values of the effective hyperfine field ( H , ) at the Fe3+ sites. As a starting approximation, since the quadrupole interaction at the ferric sites is considerably smaller than the Zeeman splitting, one can estimate H , from the overall spectral splitting. In this case = 14.67 mm/s from the least-squares fit implying HCff = 454 kG. More accurate values for H , can be obtained from the I = 1 / 2 ground-state Zeeman splitting (there is no quadrupole splitting for the ground state) as deduced from the separation of transition pairs terminating in the same m I sublevel of the I = 3/2 excited state. The spectral pairs involved are 3' and 5' and 2' and 4' for which least-squares fitting gives and A2t-4~equal to 5.44 and 5.51 mm/s, respectively, corresponding to H , = 457 and 463 kG for a ground-state splitting of 0.1188 mm s-I T-' or thus an average value of H , = 460 kG at ferric sites at 20 K. An easy check on our choice of appropriate pairs comes from the fact that the area of such pairs should equal" one-fourth of the total spectral absorption intensity for a thin absorber. For the least-squares fit of Figure 8 and the pairs we have chosen, this is precisely the case. Imbert et al.'] have examined the Mossbauer spectra of this material in detail. From their analysis of the Fe(I1) subspectrum, they concluded that Fe(I1) ions existed in an axial crystal field, in agreement with the structural results and that Fe(I1) exhibits a strong magnetic anisotropy, again consistent with expectations based on the structure. The Mossbauer spectrum of Fe2F5.2H20at 4.2 K, H = 0, is very similar to that at 20 K except that the effective field A31-j,

I I " " " ' L .I d 4 4 0 2 4 6 VELOClTY(mW1

I)

Figure 9. Mossbauer spectra of Fe2Fj.2H20at 4.2 K in longitudinal applied fields of 0, 5, and 60 kG.

at the ferric sites has increased, Le., H , = 532 kG. This result agrees favorably with that of Imbert et al., who found H , = 538 kG. Based on that value and a comparison to iron fluorides of known structure, they concluded that the geometry at Fe(II1) was Fe"'F5(H20), as in the complex Fe2F5.7H20. However, the crystal structure of Fe2Fj.2H20 clearly demonstrates Fe111F6coordination.'* Spectra in small and large applied fields show degrees of broadening and no obvious evidence of field-induced phase transitions such as are observed from susceptibility measurements. Some interesting observations can be made from the field dependence of the powder spectra. First of all, one sees a positive hyperfine field at the ferric sites. That is, at 4.2 K the value of H , at the ferric sites is 549 kG for HO= 60 kG or an increase of -17 kG relative to Ho = 0. A comparable effect is observed for FeF3-3H20for which H , increases from 446 to 456 kG on increasing the applied field from 0 to 60 kG. In Figure 9 it should also be noted that transitions 2' and 5' (the AmI = 0 transitions) of the ferric subspectrum have intensified relative to 1' and 6' indicating rotation of the spin moments of the crystallites and thus their magnetization to a direction normal to that of the applied field. Powder samples of FeF,.3H20 also show a similar field-de-

Kinetics of the Trihydroxovanadyl(1V) Ion pendent behavior. For sufficiently large applied fields, it is possible for canting of the Fe2+and Fe3+ sublattice moments to occur. However, the high critical temperature of the present system and concomitant strong intersublattice exchange interaction suggest that such canting will be negligible and not observable in the powder Mossbauer spectra. To conclude it is worthwhile to compare some of our low-temperature results with those for the higher hydrate Fe2F5.7H20. This material orders at T < 3 K and the details of its zero-field Mossbauer spectrum at 1.3 K have been discussed.26 In contrast to the dihydrate there appear to be two slightly different but nearly equally populated ferric sites for which H, = 561 and 573 kG. The effective field at the ferrous sites of Fe2F5.7H20is 146 kG at 1.3 K and is much larger than that observed (41 kG) for the ferrous sites of the dihydrate. The considerable difference in the critical temperatures of these materials, 48.0 K for the dihydrate vs.